Hello Matt,
Thank you very much! 

Sifan
From: Matt Blackwell [blackwel@fas.harvard.edu]
Sent: Monday, April 30, 2012 3:47 PM
To: Zhou, Sifan
Cc: cem-bounces@lists.gking.harvard.edu; cem@lists.gking.harvard.edu
Subject: Re: [cem] how to incorporate sampling weight with cem_weights?

Hi Sifan, 

There are many different ways to incorporate weights in Stata routines. For the "regress" command, "iweight" and "aweight" give the same result, but, for example, there is no "aweight" option for "logit".   It isn't always clear how Stata uses each of these weights. 

You are right that you can simply multiply the two weights together and use "pweights". 

Hope that helps!

Cheers,
matt.

On Sunday, April 22, 2012 at 10:12 PM, Zhou, Sifan wrote:

Thank you, Ariel.
I was referring to "cem: Coarsened exact matching in Stata"(by Matthew Blackwell et.al, The Stata Journal 2009), in which iweight is used usually and aweight is used when we have multiple imputation. Or has that been changed since 2009? If so, could you please kindly offer me a reference? I hope I can find out how these weights are used in estimations.
And I did find two weighting method behave differently. For the same cem_weights, using pweight or iweight gives different results. I will also try aweight.
Sifan

________________________________________
Sent: Sunday, April 22, 2012 4:42 PM
Subject: Re: [cem] how to incorporate sampling weight with cem_weights?

I believe an "aweight" is used for cem_weight in Stata, not "iweight". I think the product of the sampling weight and cem_weight should suffice, assuming the aweight and pweight don't behave differently...

Ariel


-----Original Message-----
From: "Zhou, Sifan" <szhou@albany.edu>
Date: Sun, 22 Apr 2012 20:20:20
Subject: [cem] how to incorporate sampling weight with cem_weights?

-
--
cem Mailing List, served by HUIT
More information on cem: http://gking.harvard.edu/cem
Cem mailing list
-
--
cem Mailing List, served by HUIT
More information on cem: http://gking.harvard.edu/cem
Cem mailing list